Thread: apply limit to sum function
Readers, Is it possible to apply the 'sum' function, but limit the value and then show the tuples of the result? For example, the table has: text1, 10 text2, 12, text3, 23 Instead of applying the 'sum' function to all values, the request is to be able to select those values that result in a sum of a target value (e.g. < 22) and then return the tuples.
On 04/27/2012 12:27 PM, e-letter wrote:
> Readers,
>
> Is it possible to apply the 'sum' function, but limit the value and
> then show the tuples of the result? For example, the table has:
>
> text1, 10
> text2, 12,
> text3, 23
>
> Instead of applying the 'sum' function to all values, the request is
> to be able to select those values that result in a sum of a target
> value (e.g.< 22) and then return the tuples.
>
Yes. Um, I think. I am trying to decipher exactly what you want. If you
just want categories that sum to more than some amount then you can use
"having":
select somelabel, sum(somevalue) from sometable group by somelabel
having sum(somevalue) > yourtargetvalue;
If you want all the original records that contribute to a sum of more
than some value, you can use the above as the basis on which to select
records having that label. There are numerous variants of this.
One way is to use the above as a subselect. Another is with common table
expressions:
with labelcount as
(select somelabel, sum(somevalue) as groupsum from sometable group
by somelabel)
select * from
sometable
join labelcount on sometable.somelabel=labelcount.somelabel
where labelcount.groupsum > yourtargetvalue;
Yet another uses windowing functions:
select * from (
select *, sum(somevalue) over (partition by somelabel) as labelcount
from sometable) as foo
where labelcount > yourtargetvalue;
Cheers,
Steve
On 27/04/2012, Steve Crawford <scrawford@pinpointresearch.com> wrote: > On 04/27/2012 12:27 PM, e-letter wrote: >> Readers, >> >> Is it possible to apply the 'sum' function, but limit the value and >> then show the tuples of the result? For example, the table has: >> >> text1, 10 >> text2, 12, >> text3, 23 >> >> Instead of applying the 'sum' function to all values, the request is >> to be able to select those values that result in a sum of a target >> value (e.g.< 22) and then return the tuples. >> > > select somelabel, sum(somevalue) from sometable group by somelabel > having sum(somevalue) > yourtargetvalue; > ... > > > select * from ( > select *, sum(somevalue) over (partition by somelabel) as labelcount > from sometable) as foo > where labelcount > yourtargetvalue; > So far, tried these methods above but produced: text1, 10 text2, 12 It seems that these queries produce a result that corresponds to _each row_ value being evaluated against the target value. The desired output is that the sum of the values in the result is to be evaluated against the target. But thanks anyway so far.
Hi On 28 April 2012 09:26, e-letter <inpost@gmail.com> wrote: > On 27/04/2012, Steve Crawford <scrawford@pinpointresearch.com> wrote: >> On 04/27/2012 12:27 PM, e-letter wrote: >>> Readers, >>> >>> Is it possible to apply the 'sum' function, but limit the value and >>> then show the tuples of the result? For example, the table has: >>> >>> text1, 10 >>> text2, 12, >>> text3, 23 >>> >>> Instead of applying the 'sum' function to all values, the request is >>> to be able to select those values that result in a sum of a target >>> value (e.g.< 22) and then return the tuples. >> >> select somelabel, sum(somevalue) from sometable group by somelabel >> having sum(somevalue) > yourtargetvalue; > ... >> select * from ( >> select *, sum(somevalue) over (partition by somelabel) as labelcount >> from sometable) as foo >> where labelcount > yourtargetvalue; > > So far, tried these methods above but produced: > > text1, 10 > text2, 12 > > It seems that these queries produce a result that corresponds to _each > row_ value being evaluated against the target value. The desired > output is that the sum of the values in the result is to be evaluated > against the target. But thanks anyway so far. I'm not entirely clear on what you want. e.g. do you want to generate all possible permutations of your values and then find the sums that are less than your target? If so, how do you decide which subset of rows to return? i.e.: 1: (text1, 10) => 10 2: (text2, 12) => 12 3: (text3, 23) => 23 4: (text1, 10), (text2, 12) => 22 5: (text1, 10), (text3, 23) => 33 6: (text1, 10), (text2, 12), (text3, 23) => 45 Only options 1 and 2 are < 22, so I assume you're looking for either (text1, 10) or (text2, 12) as an answer given this example data? How do you choose which one? Or do you just want to take the data in the order given and add the values until the total goes above 22 and then return the rows you had just before going over the target value? Or something else? -- Michael Wood <esiotrot@gmail.com>
On 29 April 2012 17:33, Michael Wood <esiotrot@gmail.com> wrote: > Hi > > On 28 April 2012 09:26, e-letter <inpost@gmail.com> wrote: >> On 27/04/2012, Steve Crawford <scrawford@pinpointresearch.com> wrote: >>> On 04/27/2012 12:27 PM, e-letter wrote: >>>> Readers, >>>> >>>> Is it possible to apply the 'sum' function, but limit the value and >>>> then show the tuples of the result? For example, the table has: >>>> >>>> text1, 10 >>>> text2, 12, >>>> text3, 23 >>>> >>>> Instead of applying the 'sum' function to all values, the request is >>>> to be able to select those values that result in a sum of a target >>>> value (e.g.< 22) and then return the tuples. >>> >>> select somelabel, sum(somevalue) from sometable group by somelabel >>> having sum(somevalue) > yourtargetvalue; >> ... >>> select * from ( >>> select *, sum(somevalue) over (partition by somelabel) as labelcount >>> from sometable) as foo >>> where labelcount > yourtargetvalue; >> >> So far, tried these methods above but produced: >> >> text1, 10 >> text2, 12 >> >> It seems that these queries produce a result that corresponds to _each >> row_ value being evaluated against the target value. The desired >> output is that the sum of the values in the result is to be evaluated >> against the target. But thanks anyway so far. > > I'm not entirely clear on what you want. > > e.g. do you want to generate all possible permutations of your values > and then find the sums that are less than your target? If so, how do > you decide which subset of rows to return? > > i.e.: > > 1: (text1, 10) => 10 > > 2: (text2, 12) => 12 > > 3: (text3, 23) => 23 > > 4: (text1, 10), (text2, 12) => 22 > > 5: (text1, 10), (text3, 23) => 33 > > 6: (text1, 10), (text2, 12), (text3, 23) => 45 Sorry, I forgot: (text2, 12), (text3, 23), but obviously that's also over the limit. > Only options 1 and 2 are < 22, so I assume you're looking for either > (text1, 10) or (text2, 12) as an answer given this example data? How > do you choose which one? > > Or do you just want to take the data in the order given and add the > values until the total goes above 22 and then return the rows you had > just before going over the target value? > > Or something else? -- Michael Wood <esiotrot@gmail.com>
On 29/04/2012, Michael Wood <esiotrot@gmail.com> wrote: > > e.g. do you want to generate all possible permutations of your values > and then find the sums that are less than your target? If so, how do > you decide which subset of rows to return? > Ideally, yes but that would make solution only via the "knapsack" combination algorithm type of method? > i.e.: > > 1: (text1, 10) => 10 > > 2: (text2, 12) => 12 > > 3: (text3, 23) => 23 > > 4: (text1, 10), (text2, 12) => 22 > > 5: (text1, 10), (text3, 23) => 33 > > 6: (text1, 10), (text2, 12), (text3, 23) => 45 > > Only options 1 and 2 are < 22, so I assume you're looking for either > (text1, 10) or (text2, 12) as an answer given this example data? How > do you choose which one? > Would like to be able to choose 'text1' and 'text2' on the basis of the sum of the values for the tuples being to the target value. > Or do you just want to take the data in the order given and add the > values until the total goes above 22 and then return the rows you had > just before going over the target value? > Yes, that would be acceptable: read values in order in the table until the sum reaches a target value, then return these rows.